Search papers, labs, and topics across Lattice.
Tsinghua AIDepartment of Data and SystemsHKUThe Research Center for DigitalThe Univer- sity of Hong KongAug 31, 2026arXiv:2608.30512This paper introduces Neural Double Q-routing, a novel approach for optimizing routing in overhead hoist transport (OHT) systems, which are critical in semiconductor manufacturing. By utilizing a shared state-action value network initialized through return-to-go regression on simulated trajectories, the method enhances information sharing and adapts to dynamic traffic conditions more effectively than traditional tabular Q-routing. The results show a significant reduction in mean completion time, achieving the best performance in larger fleet settings while maintaining task completion rates comparable to existing methods.
Neural Double Q-routing slashes mean completion times by up to 8.8% in complex industrial environments, outperforming traditional methods.
Large-scale industrial robot fleets share constrained physical infrastructure, making vehicle travel times dependent on safety separation, intersection access, downstream blocking, and station contention. We study this problem in overhead hoist transport (OHT) systems, a representative ceiling-mounted material-handling system used in semiconductor fabs. Static shortest-path routing cannot account for these time-varying traffic costs, whereas tabular Q-routing adapts online but learns each destination--node--action value independently, limiting information sharing across sparsely visited routing contexts and making startup behavior sensitive to inaccurate value estimates. We propose Neural Double Q-routing, which replaces destination-indexed tables with a shared state--action value network. The network is warm-started through return-to-go regression on mixed simulator-generated routing trajectories and then refined online using Double-Q updates, local congestion correction, and event-stratified structured replay. Across nine matched fleet-size--arrival-rate settings with 100, 150, and 200 OHTs, the proposed framework reduces mean completion time relative to tabular Double Q-routing by $0.8\%$--$8.8\%$. It achieves the lowest mean completion time among all compared methods in the six 150- and 200-OHT settings, whereas Dijkstra remains best in the three 100-OHT settings. Completed-task counts remain within $1\%$ of tabular Double Q-routing in eight of nine settings, and 95th-percentile completion time decreases in eight settings. In two matched startup scenarios, offline initialization increases the number of completed tasks by up to $23\%$ and reduces tail completion time by up to $15\%$.
